India and Global South AI Governance

Quick Assessment

Key Links

Overview

India and Global South AI GovernanceOrganizationIndia and Global South AI GovernanceA comprehensive overview of India's development-centric AI governance approach and the Global South's participation in reshaping global AI policy, centered on the 2026 India AI Impact Summit and fr... refers to the coordinated effort by India and a coalition of developing nations — spanning Africa, Latin America, South Asia, and Southeast Asia — to reshape global AI regulation around development priorities rather than the safety-first or liability-focused frameworks championed by the United States and the European Union. Civil society researchers and policy advocates, including those associated with Research ICT Africa's Africa Just AI Project, argue that countries representing the majority of the world's population have been largely absent from the rule-making processes governing AI, and that a governance architecture built without their participation is unlikely to produce outcomes that serve their populations.¹

India has sought to position itself as a leading convenor and strategic bridge for this agenda. Its approach, articulated through NITI Aayog's foundational #AIforAll strategy, the IndiaAI Mission, and the November 2025 India AI Governance Guidelines, emphasizes incentivizing innovation and building domestic AI infrastructure rather than imposing prescriptive regulation comparable to the EU AI ActPolicyEU AI ActComprehensive overview of the EU AI Act's risk-based regulatory framework, particularly its two-tier approach to foundation models that distinguishes between standard and systemic risk AI systems. ...Quality: 55/100. India's governance philosophy rests on seven principles — Trust, People First, Innovation over Restraint, Fairness and Equity, Accountability, Understandable by Design, and Safety, Resilience and Sustainability — and explicitly frames economic inclusion and sovereignty as inseparable from responsible AI development.²

The broader Global South governance agenda extends well beyond India. Brazil's Bill 2338/2023 and its 2025 BRICS presidency declaration on AI, the African Union's Continental AI Strategy 2024 and nascent African AI Council concept, ASEAN's Guide on AI Governance and Ethics 2024, Kenya's National AI Strategy (2025–2030), and emerging positions from Indonesia, Nigeria, and Vietnam collectively represent a significant reorientation of AI governance and policyCruxAI Governance and PolicyComprehensive analysis of AI governance mechanisms estimating 30-50% probability of meaningful regulation by 2027 and 5-25% x-risk reduction potential through coordinated international approaches. ...Quality: 66/100 away from a largely transatlantic conversation. Researchers and policymakers have argued these actors matter not only because of demographic weight — the Global South contains the majority of the world's AI users and workers — but because of their role in global data generation, the strategic dynamics of the US–China technology competition, and the expanding network of national AI Safety Institutes that requires their participation to be globally legitimate.³

History and Background

NITI Aayog and the #AIforAll Foundation

India's formal engagement with AI governance began with NITI Aayog's National Strategy for Artificial Intelligence, published in June 2018, which established the foundational concept of #AIforAll — framing India's AI leadership as oriented toward inclusive development rather than narrow commercial or military advantage. This strategy positioned AI across priority sectors including agriculture, healthcare, education, and smart mobility, emphasizing India's potential to demonstrate that AI adoption at scale could serve social equity goals.⁴

The strategy also reflected a pragmatic recognition of India's structural advantages: a large and linguistically diverse population generating volumes of data, a substantial software engineering talent pool, and existing Digital Public Infrastructure (DPI) in the form of Aadhaar (biometric identity), UPI (payments), and DigiLocker (document storage). These systems, collectively called India Stack, provided a governance-adjacent infrastructure that few other developing countries had assembled at comparable scale. As of March 2026, Aadhaar had generated approximately 1.44 billion unique identity numbers with over 17,080 crore authentications performed; UPI processed 172 billion transactions in 2024 with a total value of approximately ₹247 lakh crore (roughly $2.9 trillion), representing 46% year-over-year transaction volume growth; and DigiLocker had reached 676 million users with more than 9,500 crore stored documents.⁵

IndiaAI Mission and Institutional Development

The IndiaAI Mission, approved by the Union Cabinet on March 7, 2024 with an outlay of approximately Rs 10,372 crore (roughly $1.25 billion) over five years, operationalized this strategy through concrete infrastructure investment.⁶ The Mission is implemented by IndiaAI, an independent business division under MeitY, and was led by CEO Abhishek Singh — an Additional Secretary at MeitY and 1995-batch IAS officer who was named to TIME100 AI 2025 for his role in steering the initiative.⁷

The Mission's compute pool, initially targeting 18,000 GPUs across seven industry partners, has since scaled to 38,231 GPUs across 14 providers, accessible to students, startups, researchers, and government users at approximately ₹65 per hour — estimated at up to 40% below prevailing commercial rates.⁸ This established a precedent for state-enabled democratization of AI infrastructure that has attracted attention from African and Latin American governments as a potential model for shared regional compute. The Mission also supports BharatGen, a multilingual generative AI initiative targeting India's 22 official languages, and the Bhashini platform for natural language processing across Indian languages — both addressing a structural imbalance in global AI training data: English comprises approximately 43.9% of Common Crawl web content despite being spoken by fewer than 20% of the world's population, while seven widely-spoken but low-resource languages including Hausa, Amharic, and Yoruba each constitute less than 0.004% of the same dataset.⁹

In March 2024, MeitY issued an initial advisory requiring platforms to obtain explicit permission before deploying under-tested AI models and to include disclaimers about potential unreliability. Following industry pushback, this advisory was withdrawn and replaced with a less restrictive version — an early indicator of the tension between India's pro-innovation orientation and precautionary impulses within Indian regulatory circles.²

The India AI Governance Guidelines (November 2025)

On November 6, 2025, MeitY released the India AI Governance Guidelines under the IndiaAI Mission, which became India's primary framework for responsible AI adoption.² The document introduces a non-prescriptive, principles-based approach rather than new legislation, favoring voluntary measures such as self-certifications and regulatory sandboxes over binding legislative mandates. The Guidelines specify a sectoral-based regulatory approach, allowing existing sectoral regulators to formulate AI-specific rules for their domains. They also recommend an institutional architecture comprising an AI Governance Group (AIGG), a Technology and Policy Expert Committee (TPEC), and an AI Safety Institute (AISI) — a structure explicitly designed to be transferable to other Global South contexts.²

The Indian government has also drafted but not yet enacted the Digital India Act, which would update the country's foundational IT law to address AI-specific concerns; the Act's draft provisions have also attracted civil liberties and freedom of expression concerns during public consultation processes. India currently lacks codified AI-specific legislation, relying instead on the IT Act, the Digital Personal Data Protection Act, and the November 2025 Guidelines as its primary governance instruments.¹⁰

India AI Impact Summit 2026

In February 2026, India hosted the India AI Impact Summit at Bharat Mandapam, New Delhi — the fourth in a global series following Bletchley Park (2023), Seoul (2024), and Paris (2025), and notably the first hosted in the Global South.¹¹ For context: the November 2023 Bletchley Park summit produced a 28-country declaration and established the first national AI Safety Institutes; the May 2024 Seoul summit secured Frontier AI Safety Commitments from 16 technology companies and pledges from 10 countries to establish national AI Safety Institutes; and the February 2025 Paris AI Action Summit produced a 60-country communiqué — with India serving as co-chair alongside France — though the United States and United Kingdom did not sign the Paris communiqué.¹²

The 2026 Summit ran from February 16–21, with Prime Minister Narendra Modi formally inaugurating proceedings. The New Delhi Declaration on AI Impact, adopted on February 19, 2026, was endorsed by 89 countries and international organisations — including Australia, France, Germany, Switzerland, Brazil, UAE, Canada, Israel, the UK, China, Japan, the USA, Russia, Saudi Arabia, Singapore, and Bangladesh — exceeding the endorsements from the 2025 Paris AI Action Summit.¹¹ Organized around three foundational pillars (People, Planet, and Progress) and seven thematic "Chakras," the Declaration established several new multilateral mechanisms: a Charter for the Democratic Diffusion of AI, a Global AI Impact Commons, a Trusted AI Commons, and an International Network of AI for Science Institutions, with an emphasis on AI education, workforce training, and energy efficiency.¹¹

The Summit showcased three sovereign AI models, secured major Big Tech infrastructure pledges, and formalized South-South working groups on compute access, multilingual data, AI literacy, and governance experimentation. It also marked India's formal launch of the IndiaAI Safety Institute (AISI) by Prime Minister Modi — announced by Minister Ashwini Vaishnaw in January 2025 and formally launched at the Summit — joining the growing international network of national AI Safety Institutes. The AISI operates with an initial budget of Rs 20 crore allocated from the Safe and Trusted Pillar of the IndiaAI Mission; it is constituted as an advisory and capacity-building body rather than a regulator, with its guidance non-binding on industry, following a hub-and-spoke model partnering with academic, research, and private sector institutions.¹¹

Key Activities and Initiatives

National Governance Frameworks

India's governance model has attracted analytical attention as a potential template for the Global South, specifically because it avoids the resource-intensive compliance infrastructure that the EU AI Act requires. A December 2024 discussion paper by Anupama Vijayakumar of the Research and Information System for Developing Countries (RIS) examining AI ethics policies from ten Global South countries — Argentina, Brazil, Chile, Egypt, India, Indonesia, Mauritius, Malaysia, Thailand, and Qatar — identified equity and Common But Differentiated Responsibilities as common ethical principles across these frameworks, and argued that an India-led development-centric model served these countries' collective interests better than adoption of EU or US frameworks. The paper identified four key concerns of Global South countries toward AI — Extractivism, Exclusion, Ethnocentrism, and Enforcement.¹³

Research on this cluster of governance approaches has noted a structural imbalance: Africa, the Pacific, and the Caribbean are described as "nearly absent from the policy discourse," with the RIS paper observing that the global discourse on AI ethics has been "dominated by entities hailing from the Global North." Researchers have estimated that a significant majority of national AI policies published since 2011 originated in high-income countries, though comprehensive cross-national counts vary across sources.¹⁴

Digital Public Infrastructure as a Governance Model

India's integration of AI with its Digital Public Infrastructure represents a distinctive governance contribution distinct from pure regulatory frameworks. Aadhaar's biometric identity system, UPI's payment network (now incorporating AI-powered voice transactions), and open health data systems like the Open Healthcare Network collectively constitute a public AI deployment architecture with implications for accountability, auditability, and equitable access that pure regulatory approaches cannot achieve.⁴

India has actively promoted this DPI model diplomatically. As of early 2026, India had signed Memoranda of Understanding with 23 countries across five continents for sharing or cooperation on India Stack components; UPI is operational in UAE, Singapore, Bhutan, Nepal, Sri Lanka, France, Mauritius, and Qatar; Africa, with six MoU partners, represents the largest regional cluster.¹⁵ This positions AI governance in India's diplomatic framing not as a set of rules but as an embedded property of shared public digital infrastructure. The Direct Benefit Transfer (DBT) system exemplifies how AI-integrated DPI can reduce corruption through traceable digital payment trails, with proponents arguing potential scalability to lower-income governance contexts.⁴

South-South Cooperation Frameworks

The Africa–Asia AI Policymaker Network, launched in Cape Town in May 2022 and institutionally housed at Research ICT Africa, provides a platform for peer learning on AI policy across the two continents, connecting policymakers from Ghana, India, Kenya, Rwanda, South Africa, and Uganda, with particular focus on regulatory experimentation in contexts underserved by Global North frameworks.⁶ The 2026 Summit formalized issue-specific South-South working groups proposing to link India's IndiaAI infrastructure with Brazil's public computing institutions and African regional hosting initiatives.

The Africa Just AI Project, run by Research ICT Africa with funding from the International Development Research Centre (IDRC) and the Swedish International Development Cooperation Agency (SIDA), represents the research dimension of this cooperation. The project emphasizes participatory processes grounded in citizens' lived realities, and its Phase 2 "Just AI Lab" interrogates AI governance through the lens of justice, inclusion, and African agency. Researchers associated with the project have argued that Africa is "largely left out of global governance debates on AI or participates from a position of relative weakness," advocating for governance frameworks that center African democratic systems, development priorities, and epistemic traditions rather than replicating models designed for high-income economies.¹

The MAP-AI initiative (Multistakeholder Approaches to Participation in AI governance), a joint project of the Centre for Communication Governance at National Law University Delhi and the Global Network Initiative, similarly works to foster multistakeholder conversations and support underrepresented voices in AI governance, with a focus on capacity building for Global South stakeholders ahead of and around major AI summits.¹⁶

Brazil, ASEAN, Africa, and Emerging Actors

Brazil has been the most legislatively active large developing economy on AI governance. Bill 2338/2023, approved by the Federal Senate in December 2024 and forwarded to the Chamber of Deputies in March 2025, represents a comprehensive risk-based AI regulatory proposal that would designate the National Data Protection Authority (ANPD) as the competent authority for AI oversight and impose penalties up to BRL 50 million or 2% of total company turnover for violations. The bill defines three risk categories: prohibited ("excessive risk") applications including social scoring systems, mass surveillance, and subliminal manipulation; a "high-risk" category covering critical infrastructure, employment decisions, access to public goods, biometric identification, and administration of justice; and a lighter-touch category for other AI systems.¹⁷ Brazil's 2025 BRICS presidency produced a declaration explicitly positioning the United Nations as the central authority for global AI governance and warning that AI risks deepening global inequalities without collective action.¹⁷

ASEAN published its Guide on AI Governance and Ethics on February 2, 2024, at the 4th ASEAN Digital Ministers' Meeting, as part of its broader Digital Master Plan. The non-binding guide provides a practical framework for organisations wishing to design, develop, and deploy traditional AI technologies, emphasizes proportionate regulation, and explicitly encourages alignment and interoperability across ASEAN jurisdictions.³

The African Union's Continental AI Strategy, endorsed by the AU Executive Council during its 45th Ordinary Session in Accra, Ghana on July 18–19, 2024, and the emerging concept of an African AI Council represent the continental governance layer, aiming to give African nations collective voice in global forums while addressing the severe compute and data infrastructure deficits that make any purely regulatory approach insufficient. Kenya's National AI Strategy (2025–2030), officially launched on March 27, 2025 following a consultative process begun in April 2024, and the Konza Technopolis initiative represent national-level implementation, while South Africa's AI Institute and its leadership of the G20 AI agenda in 2025 gave the continent an additional seat at high-table discussions.³

Indonesia, Nigeria, and Vietnam are all developing AI policy positions with increasing institutional specificity. Indonesia released its first comprehensive National AI Roadmap whitepaper in August 2025, targeting 100,000 AI talents annually with a vision extending to 2045 and plans for a National Data and AI Ethics Council. Nigeria published its National AI Strategy in August 2024, articulating five pillars including responsible AI governance and proposing a National AI Steering Committee; in 2025, Nigeria introduced a National Artificial Intelligence Commission (Establishment) Bill in its legislature. Vietnam has committed to enacting a dedicated AI Law by end of 2025, framing AI as infrastructure as fundamental as electricity, with a risk-based framework targeting every citizen with a "personal digital assistant."¹⁸ Their collective significance is demographic and strategic: they represent hundreds of millions of AI users, major data generation nodes, and important actors in any attempt to build globally legitimate AI governance.

The 'AI for Development' Agenda vs. AI Safety

A critical conceptual distinction runs through Global South AI governance: the AI for Development framing treats AI primarily as a tool for economic growth, poverty reduction, agricultural productivity, and public service delivery — applications where risks of over-restriction are considered more serious than risks of under-regulation. This contrasts with the AI Safety framing that has dominated discourse in the United States and United Kingdom, which prioritizes frontier model risks, existential hazards, and alignment failures as the central governance challenges.¹⁹

The AI governanceCruxAI Governance and PolicyComprehensive analysis of AI governance mechanisms estimating 30-50% probability of meaningful regulation by 2027 and 5-25% x-risk reduction potential through coordinated international approaches. ...Quality: 66/100 implications of this divide are significant. Global South actors have generally supported the expansion of the international AI Safety Institute network, including India's own AISI, while simultaneously arguing that the safety agenda has been defined primarily around concerns most relevant to frontier AI developers concentrated in a handful of wealthy countries. Research from the MAP-AI initiative and civil society organizations associated with the Global Digital Justice Forum advocates for "lifecycle accountability" approaches that address immediate, concrete harms — discrimination, surveillance misuse, disinformation — as the stated safety priority for Global South contexts, in contrast to the frontier-risk framing that characterizes Bletchley-summit-era AI governance.¹⁶

Funding

The financial architecture of India's Global South AI governance leadership combines public investment, Big Tech partnerships, and multilateral channels:

Total investment announced at the 2026 Summit was reported by Fortune as exceeding $200 billion, aggregating commitments including Reliance Industries ($110B over 7 years), Adani Enterprises ($100B by 2035), and Big Tech pledges above; this figure represents cumulative varied commitments and timelines across multiple actors rather than a single audited total.²⁰

Key People

• Narendra Modi (Prime Minister of India): Frames India's AI agenda primarily around economic growth and Big Tech partnership. Co-chaired the 2025 Paris AI Action Summit with French President Macron; inaugurated the 2026 India AI Impact Summit. Civil society observers, including analysts at Data Privacy Brasil, have characterized his government's approach as centered on government–Big Tech partnerships and industrial policy, with limited structural mechanisms for civil society participation.¹¹
• Abhishek Singh (CEO, IndiaAI Mission): A 1995-batch IAS officer and Additional Secretary at MeitY, Singh served as the operational head of the IndiaAI Mission, overseeing the establishment of the public compute pool and other Mission pillars. He was named to TIME100 AI 2025 for his work steering the initiative.⁷
• Nandan Nilekani: Architect of Aadhaar and India Stack; associated with the DPI-as-governance model that underlies India's distinctive approach to AI deployment accountability. His influence on the framing of India's digital public infrastructure is widely referenced in policy discussions. Available public sources do not confirm a current formal advisory role with MeitY or the IndiaAI Mission; he serves as non-executive chairman of Infosys and has written and spoken extensively about DPI-based AI strategy.⁴
• Global South Alliance (GSA): Civil society coalition of 26 organizations, founded in 2022 by Aapti Institute (India), Data Privacy Brasil, and Paradigm Initiative (Pan-African); founding members also include the Centre for Communication Governance at National Law University Delhi, KICTANET (Kenya), CIPESA, Derechos Digitales, Research ICT Africa, and Pollicy, spanning Latin America, Africa, and Asia. Approximately 15 of the 26 organizations participated in person at the 2026 Summit. The GSA submitted a pre-Summit letter advocating for stronger and more institutionalized participation and a rights-based approach on the future of work and contextualized AI governance; according to post-Summit analysis by Data Privacy Brasil, the letter's recommendations were not reflected in the Summit's approach.²¹
• Research ICT Africa (Africa Just AI Project): Brought participatory governance evidence from the Africa Just AI research program to the Summit, representing civil society perspectives grounded in African deployment contexts and advocating for governance frameworks centered on African agency and justice.¹

Criticism

Assessments of India's Global South AI governance leadership — including from sympathetic researchers — identify several significant shortcomings that complicate the more optimistic framing of India as a development-centric alternative to Western governance models.

Missed institutionalization at the Summit: Despite raising global expectations as the first AI summit hosted in the Global South, the 2026 India AI Impact Summit produced a Leaders' Declaration with no binding commitments and limited clarity on implementation pathways. Analysts at Data Privacy Brasil, reviewing the Summit outcomes from a Global South civil society perspective, concluded that the event reproduced patterns from prior summits — government–Big Tech showcasing — and maintained "a fragmented approach, with discussions continuing to prioritize innovation and future risks over the pressing structural needs of the Global South."¹⁶

Corporate-government dominance over civil society: Despite rhetoric about multistakeholderism, no structured mechanisms ensured meaningful participation by Global South civil society or academic institutions in the negotiated outcomes. The Summit's primary deliverables centered on Big Tech investment pledges aligned with Modi's economic growth agenda, and the Global South Alliance's calls for institutionalized participation were not reflected in the Summit's approach.¹⁶

Regulatory gaps and enforcement deficits: India currently lacks codified AI legislation, binding regulatory frameworks, a dedicated AI regulatory body, or a legal definition of AI. Broader Global South governance suffers from weak data protection regimes, compute and data concentration in external cloud providers, and institutional readiness deficits that make voluntary governance guidelines insufficient.¹⁰ The March 2024 advisory withdrawal following industry pressure illustrates the political difficulty of maintaining even soft regulatory positions against commercial interests.

The 'magical cure' problem: The RIS discussion paper examining AI governance across Global South countries identifies a pattern in some national AI strategies of treating AI as a solution to development challenges without building the localized governance, digital literacy, or R&D infrastructure necessary to deploy it responsibly. This risks positioning developing countries as passive consumers and testing grounds for technologies designed elsewhere.¹⁴

Terminology and homogenization: Multiple researchers note that "Global South" in AI discourse risks implying false homogeneity across vastly different countries — conflating India's 1.4 billion people and substantial economy with much smaller, lower-capacity states — in ways that obscure the actual diversity of governance needs and capabilities.¹³

The Trump administration's opposition: At the India AI Impact Summit, White House technology adviser Michael KratsiosPersonMichael KratsiosDirector of the White House Office of Science and Technology Policy under the Trump administration. Key architect of the 2026 National AI Legislative Framework. declared that the United States "totally rejects global governance of AI," arguing that AI policy "must be local" and warning that international forums risk creating "bureaucratic management and centralisation" under the guise of safety. The US urged nations to adopt a "sovereign AI model" built on the American technology stack, highlighting the structural difficulty of any India-led effort to build binding multilateral consensus when the primary AI superpower explicitly opts out.²²

Workforce risks: India's planning commission has warned that the technology services sector could lose 1.5 to 2 million jobs to automation by 2031, while potentially creating up to 4 million new roles — a net positive projection that depends on successful reskilling. Indian IT companies including TCS reduced their collective workforce by more than 60,000 employees in 2025 partly due to automation, with TCS alone cutting approximately 12,200 positions in July 2025.²³ India's annual engineering graduate output — approximately 1.5 million annually according to All India Council for Technical Education (AICTE) data — represents both a strategic asset and a population potentially affected by the AI productivity transitions India is actively accelerating.²⁴

Key Uncertainties

• Whether the South-South working groups formalized at the 2026 Summit will achieve concrete coordination outcomes, or dissipate as working groups from prior summits have
• Whether India's AISI will meaningfully bridge the "AI for Development" and "AI Safety" agendas, or remain focused on governance rather than technical safety research
• Whether Brazil's Bill 2338/2023 will be enacted by the Chamber of Deputies and, if so, whether it will provide a legislative template for other Latin American and African countries
• Whether the AU Continental AI Strategy 2024 and African AI Council concept will achieve sufficient institutional backing to function as a real counterweight to Global North standard-setting bodies
• The extent to which India's DPI model is genuinely transferable to lower-capacity states given the unique political, institutional, and demographic conditions that enabled its construction